Rationale and Design of the Leipzig (LIFE) Heart Study: Phenotyping and Cardiovascular Characteristics of Patients with Coronary Artery Disease

Objective We established the Leipzig (LIFE) Heart Study, a biobank and database of patients with different stages of coronary artery disease (CAD) for studies of clinical, metabolic, cellular and genetic factors of cardiovascular diseases. Design The Leipzig (LIFE) Heart Study (NCT00497887) is an ongoing observational angiographic study including subjects with different entities of CAD. Cohort 1, patients undergoing first-time diagnostic coronary angiography due to suspected stable CAD with previously untreated coronary arteries. Cohort 2, patients with acute myocardial infarction (MI) requiring percutaneous revascularization. Cohort 3, patients with known left main coronary artery disease (LMCAD). Results We present preliminary results of demographics and phenotyping based on a 4-years analysis of a total of 3,165 subjects. Cohort 1 (n = 2,274) shows the typical distribution of elective coronary angiography cohorts with 43% cases with obstructive CAD and 37% normal angiograms. Cohorts 2 and 3 consist of 590 and 301 subjects, respectively, adding patients with severe forms of CAD. The suitability of the database and biobank to perform association studies was confirmed by replication of the CAD susceptibility locus on chromosome 9p21 (OR per allele: 1.55 (any CAD), 1.54 (MI), 1.74 (LMCAD), p<10−6, respectively). A novel finding was that patients with LMCAD had a stronger association with 9p21 than patients with obstructive CAD without LMCAD (OR 1.22, p = 0.042). In contrast, 9p21 did not associate with myocardial infarction in excess of stable CAD. Conclusion The Leipzig (LIFE) Heart Study provides a basis to identify molecular targets related to atherogenesis and associated metabolic disorders. The study may contribute to an improvement of individual prediction, prevention, and treatment of CAD.


Introduction
Coronary artery disease (CAD) and its complications such as myocardial infarction and congestive heart failure are projected to remain the leading cause of mortality in the world [1]. CAD is caused by a complex pattern of interaction of genetic factors and life-style. However, individual CAD susceptibility is not well understood. Combined assessment of traditional risk factors, clinical phenotypes, genetic information as well as proteomic and metabolic data is a promising strategy for the assessment of the individual cardiovascular risk -the premise of 'personalized medicine' [2].
A significant source of limitation in many previous genetic studies of CAD has been from imprecise phenotypic character-ization of cases and controls [3,4]. Coronary lesions show a wide range of clinical manifestations, from asymptomatic to stable symptomatic disease with stress-induced angina to the point of acute coronary syndrome, myocardial infarction and even cardiac death. This contributes substantially to the heterogeneity of cohorts that have been assessed to date and many of them are a medley of subjects with stable angina, acute coronary syndrome and with a history of myocardial infarction or coronary revascularization. A detailed and highly standardized phenotypic characterization might therefore contribute to a more valid classification of cases and controls. Moreover, a precise assessment of confounders such as life-style factors and concomitant disease might allow for a more appropriate adjustment for these confounders. The latter will also be important for validation of novel biomarkers using e.g. transcriptomic, proteomic or metabolomic analysis [5,6].
The Leipzig (LIFE) Heart Study represents a biobank and data base with detailed cardiovascular, metabolic and biochemical characterization of patients with different entities of atherosclerotic coronary disease. These include subjects undergoing first-time coronary angiography due to suspected CAD but previously unknown coronary status (cohort 1), subjects with acute myocardial infarction as first manifestation of CAD (cohort 2) and subjects with left main coronary artery disease (cohort 3). The suitability of the database and biobank to perform association analyses will be supported by assessing the CAD susceptibility locus on chromosome 9p21.

Methods
A detailed description of methods is provided in the online supplement (Data S1).

Overview
The Leipzig (LIFE) Heart Study is designed as observational study to evaluate biochemical and molecular biomarkers and their ability to assess the presence and severity of CAD in symptomatic subjects (cross-sectional) and to predict the future course of disease (longitudinal). Follow-up at 5-year intervals will provide prospective information about major cardiac clinical events of the initial study subjects (cardiovascular death, myocardial infarction/reinfarction, coronary revascularization). The study meets the ethical standards of the Declaration of Helsinki. It has been approved by the Ethics Committee of the Medical Faculty of the University Leipzig, Germany (Reg. No 276-2005) and is registered with ClinicalTrials.gov (NCT00497887). Written informed consent has been obtained from all participants involved in the study.

Study cohorts
Cohort 1: Subjects with suspected CAD. Patients included in cohort 1 are referred to coronary angiography by an outpatient cardiologist. Coronary angiography was indicated by clinical symptoms and non-invasive testing. Patients with any previous coronary revascularization in form of percutaneous coronary intervention (PCI) or coronary artery bypass graft (CABG) are excluded to obtain subjects with coronary first-time events and untreated coronary arteries (Table 1).
Cohort 2: Subjects with myocardial infarction. Patients included in cohort 2 required primary or rescue PCI caused by acute myocardial infarction. Symptom onset and interval to revascularization is limited to a maximum of 60 and 36 hours before enrollment and blood taking, respectively. The rationale is to evaluate circulating biomarkers of the early postinfarction phase and their ability to predict the clinical outcome and risk for future coronary events (Table 1).
Cohort 3: Subjects with left main coronary artery disease (LMCAD). Patients included in cohort 3 have known CAD and significant LMCAD with luminal reduction $50% ostial, midshaft or at bifurcation. Subjects may be treated conservative or had required revascularization in history (Table 1).

Data collection
Data collection in the Leipzig (LIFE) Heart Study reverts to standardized, validated instruments and procedures which have been proved of value in epidemiologic and clinical studies. Medical and family history, life style habits and medication are surveyed in a standardized interview. Subjects with suspected CAD pass an extended standardized study program including anthropometric measurements, blood pressure measurement, electrocardiogram, exercise test, echocardiography ( Figure S1, Table S1) and ultrasound of carotid ( Figure S2) and peripheral arteries besides coronary angiography. Therefore, atherosclerotic burden is estimated in the three mostly affected arterial beds: peripheral, carotid and coronary arteries. Coronary angiography is performed following the standards of the institution. Gensini score index is calculated and patients are classified in sets of subjects with normal angiogram (no CAD), subjects with wall irregularities but ,50% luminal reduction (CAD,50%) and subjects with obstructive CAD defined as at least one stenosis of 50% or more in any coronary vessel (CAD$50%) ( Figure S3). Significant LMCAD was defined as visual luminal reduction $50% of the left main trunk ( Figure S4).

Laboratory
Blood of subjects with suspected CAD (cohort 1) is drawn before invasive diagnostic, hence unaffected by revascularization procedures and escalation of pharmacological therapy. Blood of subjects with acute myocardial infarction (cohort 2) is drawn within 6 to 36 hours after the interventional revascularization. Blood of patients with LMCAD (cohort 3) is drawn irrespective of earlier coronary procedures. Multiple aliquots of serum, plasma, whole blood, DNA, RNA and peripheral mononuclear blood cells are stored at 280uC or liquid nitrogen for further analysis. Clinical chemistry and a set of metabolic, cardiac and inflammatory marker are measured on the day of blood sampling using an automated Roche Modular analysis system (Roche Diagnostics, Mannheim, Germany). Single nucleotide polymorphisms (SNP) rs10757274, rs2383206, rs2383207 and rs10757278 are genotyped for replication studies of the CAD-susceptibility locus on chr9p21.

Statistics
The sample size calculation to estimate the suitability of our study for genetic studies is given in the Table S2. SPSS version 16.0 was used for statistical analyses. Categorical data are expressed as numbers and percentages and were compared using the x 2 or Fisher's exact test, as appropriate. Continuous variables are expressed as the arithmetic mean 6 standard deviation for normally distributed variables and as geometric mean 6 standard deviation for non-normally distributed variables. They were compared using a one-way ANOVA and Kruskal-Wallis test with post-hoc test, respectively. P-values,0.05 were considered statistically significant. Genetic association analyses were performed using logistic regression techniques. Associations of cardiologic endpoints with single SNPs and the GGGG haplotype were adjusted for age, sex, diabetes and cigarette smoking packyears. CAD phenotypes (except myocardial infarction) were also adjusted for HDL and LDL. Additive models were calculated throughout. Calculations were performed using the statistical software package R (www.r-project.org).

Description of study cohorts
Recruitment of subjects started in December 2006 and is ongoing. The current results are based on 3.165 subjects in 3 cohorts with different entities of CAD (Table 2).

Cohort 1: Suspected coronary artery disease
Cohort 1 consists of 1,457 men and 817 women, who were referred for diagnostic coronary angiography due to suspected coronary artery disease. Completeness of phenotyping was high for blood sampling, laboratory analyses, interview and ECG (100%, respectively), echocardiography (99%), carotid ultrasound (96%) and ABI (95%). An exercise stress test was completed in 56% of the study participants while remaining patients had cardiologic or orthopedic disabilities. Coronary angiography was performed in 97% of study patients (indication for coronary angiography was revised in n = 100 subjects after study enrollment). Angiographically detectable atherosclerosis was present in the coronary tree of 71% of men and 48% of women. The remaining subjects were free of angiographically visible CAD. 49% of males and 29% of females had luminal reduction of $50% in at least one major coronary artery, resulting in assignment to percutaneous coronary intervention or coronary artery bypass grafting in 38% of males and 22% of females. 32% of men and 18% of women had a multi-vessel disease. Gensini scores of the male and female subsets with CAD$50% were 40632 and 32628, respectively. Distributions of angiographic findings are shown in Figure 1 as age strata in 10-years intervals. All age strata comprise subjects with various stages of CAD as well as subjects with angiographically normal coronary arteries.

Cohort 2: Acute myocardial infarction
Cohort 2 consists of 431 men and 159 women with acute myocardial infarction, which were included 6 to 36 hours after PCI. The mean time from PCI to inclusion was 2068 hours. 70% had an ST-elevation myocardial infarction (CK 36634 U/l) and remaining patients occurred with non-ST-elevation myocardial infarction (CK 15614 U/l). Myocardial infarction resulted in a reduction of left ventricular ejection fraction ,50% in 48% of STEMI cases and 34% of NSTEMI cases. All subjects with myocardial infarction had an identifiable culprit lesion, in 34% it was localized in the right coronary artery, 48% in the left anterior descending, and 18% in the circumflex or affiliated branches. 52% had a single-vessel-disease while 48% had a multi-vessel disease yet with further significant stenosis besides the culprit lesion. Gensini scores were 40629 in NSTEMI cases and 45628 in STEMI cases.

Cohort 3: Left main coronary artery disease
Cohort 3 consists of 242 men and 59 women with known left main coronary artery disease ($50% stenosis). Patients were recruited at an average of 6 years after diagnosis (range 0-37 years). Therefore, most of these individuals had already undergone coronary revascularization by PCI (35% ), bypass graft (30%) or both (19%). Average age at first diagnosis of CAD was 63611 years. Interestingly, 46% of these patients reported a myocardial infarction in their recollected history. According to retrospective angiograms mean value of luminal stenosis was 70617% (range 50-100%).

Demographic and cardiovascular risk factors
Demographic and clinical characteristics of study subjects are summarized for each cohort and for men and women, separately (Table 3 and 4). As expected from the earlier onset of CAD, men included in the study were significantly younger than women (62.0611.6 and 64.7611.2 years, respectively). Study participants were obese with mean body-mass index .28 kg/m 2 in all three cohorts. One third of study subjects were categorized as diabetic in both genders. In cohort 1, 21% of men and 11% of women were current smoker and 52% and 16% had a history of smoking, respectively. In cohort 2 the proportion of current smokers was much higher (49% and 36% of men and women, respectively). Comparison of subjects with angiographically normal and diseased coronary arteries showed an expected distribution of traditional risk factors for both genders. CAD-patients were older, smoked more, were more often diabetic, had lower levels of HDL, higher levels of LDL and higher levels of C-reactive protein (Table 3 and 4). Interestingly, the majority of subjects in cohort 1 were already treated with acetylsalicylic acid (52%), renin angiotensin aldosteron system inhibitors (69%), betablocker (58%), Calcium antagonists (24%), diuretics (39%) and statins (35%) before invasive coronary diagnostics.

Replication study of CAD susceptibility locus on Chromosome 9p21
To assess the suitability for performing genetic association studies, we finally tested for an association of CAD severity with the chromosome 9p21 genotype. The latter had been identified as the most robust locus associated with coronary artery disease and myocardial infarction in recent genome-wide association studies [7][8][9][10]. Replication of the chromosome 9p21 locus has been suggested as quality control for further genetic studies of cardiovascular disease [11]. We genotyped SNPs rs10757274, rs2383206, rs2383207 and rs10757278 of the CAD-risk haplotypic block on chromosome 9p21 (call rates 99%, 97%, 99% and 98%, respectively). The risk allele (G) frequencies were 50%, 52%, 53% and 49%, respectively, and were in strong linkage disequilibrium (minimum of pairwise R 2 = 0.83, Figure S5). G-alleles of all four SNPs were associated with obstructive CAD, myocardial infarction and LMCAD when comparing diseased subjects to those with normal angiograms (Figure 2). The odds ratios of the riskhaplotype were between 1.54 to 1.74 per allele (p-values 7.1610 28 to 6.9610 27 ) for all these forms of CAD using a logistic regression analysis under the assumption of an additive model and after adjustment for the traditional risk factors (Table 5). To assess the effect of the 9p21 locus between different forms of CAD, we next performed associations within CAD cohorts. Comparing subjects with stable obstructive CAD to subjects with myocardial infarction, no association with 9p21 variants was seen (OR 1.00, CI 0.87-1.15). In contrast, patients with severe forms of CAD such as LMCAD had a stronger association with 9p21 than patients with obstructive CAD without LMCAD (OR 1.22, CI 1.01-1.48), even though it must be noted that these results were at the margin of statistical significance (p = 0.042; Figure 2, Table 6).

Discussion
Here, we present the study design and data of 3.165 subjects enrolled in the ongoing Leipzig (LIFE) Heart Study. The study includes subjects with different severity and courses of coronary artery disease. CAD was prevalent in 78% of males and 57% of females, while the remaining subjects were free of angiographically detectable CAD. Patients with obstructive CAD had the expected higher level of conventional risk factors and confirmation of the chromosome 9p21 risk locus indicates excellent power for genetic association studies. Moreover, different entities of CAD included in the study allowed detailed analyses within CAD populations.
Cohort 1 of the Leipzig (LIFE) Heart Study presents a typical cohort of patients undergoing elective coronary angiography for suspected CAD. In this cohort, 43% of individuals had obstructive CAD while 37% were free of angiographically visible CAD. These data are comparable with the NCDR CathPCI Registry of 397.954 individuals [12], where 38% had obstructive CAD and 39% no CAD (Figure 1). The selection of cases and controls in genetic studies of CAD is a matter of ongoing discussion. Recent genome-wide association studies of CAD and MI included patients based on retrospective data from CAD/MI-registries, retrospective selection of premature familial CAD/MI cases and subjects from several cross sectional studies with angiographically determined CAD [7][8][9][10]13]. The controls used in some of these studies were recruited from study cohorts with non-cardiovascular disease such as neurological, psychiatric, infectious disease and cancer or subjects from population-based studies or healthy blood donors. Only few studies used angiographically determined controls without CAD for some of their replications [7,8].
The 'ideal' control would be a .90 years old individual with no prior history of CAD and completely normal coronary arteries. 4 However, this design appears unrealistic and would not permit the collection of sufficiently large cohorts needed for detection of small effects or even gene-interactions. Therefore, Luo et al. presented apparently less stringent criteria for phenotype categorization [3]: Controls should be characterized by normal coronary arteries by conventional coronary angiography or multidetector computer tomography (CT), no family history of CAD, no history of cerebrovascular or peripheral artery disease and age of controls much greater than cases (10-20 years). In contrast, cases for coronary artery disease and myocardial infarction, respectively, should present with at least one angiographic stenosis $70% in a major epicardial artery, family history of CAD, no history of smoking, no diabetes, normal or low LDL-cholesterol, high HDLcholesterol and normal C reactive protein. However, today no study exists that meets these criteria. From the total of 2274 subjects included in the untreated and unaffected cohort 1, only 1.5% and 8.5% fulfilled the complete set of criteria suggested by Luo et al for CAD cases and controls, respectively. A recent study has investigated the impact of heterogeneity of phenotype definition of CAD on genetic association studies [14]. Remarkably, these investigators found that differences in phenotype definition make only a small contribution to between study heterogeneity. As a limitation, these results were mostly based on studies of candidate genes, which did not play a role in recent genome-wide association studies of CAD. The authors conclude that even though utility of a clear phenotype definition of CAD might be limited for primary analyses, it might be helpful for secondary analyses after an association with CAD has been established [14].
The mode of recruitment of including only subjects referred to coronary angiography for suspected CAD in cohort 1 is unique to the Leipzig (LIFE) Heart Study and has the advantage of simultaneous collection of cases and controls under equal conditions. Other large angiographic studies have broader BMI, body-mass index in kg/m 2 ; SBP/DBP, systolic and diastolic blood pressure in mmHg; Total-C/HDL-C/LDL-C, cholesterol in mmol/l; hsCRP, high sensitive C-reactive protein in mg/l; RAAS, renin-angiotensin-aldosterone-system; ASA, acetylsalicylic acid. Values are given as arithmetic mean 6 standard deviation for normally distributed parameters, geometric mean 6 standard deviation for non normal distributed parameters or as percent of subjects for categorical items. Statistic differences between no CAD and CAD $50 were calculated using Students t-test/Kruskal-Walis test for scaled items and Chi-square test for categorical items. a including 61 men without angiography. doi:10.1371/journal.pone.0029070.t003  inclusion criteria and also recruit patients with non-atherosclerotic indications for coronary angiography such as valvular disease [15].
Only few studies perform combined examinations of the three mostly affected regions of atherosclerosis development such as coronary, carotid and peripheral arteries. These studies use computed tomography of coronary calcium as a surrogate parameter for coronary atherosclerosis [16][17][18]. To our knowledge, the Leipzig (LIFE) Heart Study is currently the only largescale cohort providing angiographic assessment of coronary arteries combined with sonographic evaluation of carotid and peripheral atherosclerosis. Another important advantage is the highly standardized preanalytical process of blood sampling prior to interventional or surgical treatment of coronary stenosis. This enables valid analytical studies of blood cell gene-expression [19] as well as profiling of mediators and biomarkers of metabolism, atherosclerotic wall pathology and myocardial function, yet unaffected by revascularization procedures and initiation of intensive pharmacological treatment. Limitations of selecting 'controls' from patients presenting with angiographically normal coronary arteries might be the presence of atherosclerotic lesions without affecting arterial lumen and patients that will develop significant lesions and coronary events in future. Angiographic luminography is not suitable to detect outward remodeling of plaques or fatty streak lesions and small fibroatheroma which often represent vulnerable plaques. In spite of these limitations, the classification of cases and controls in the Leipzig (LIFE) Heart Study according to angiographic findings led to an initial cohort with sufficient power to confirm the chromosome 9p21 locus for CAD and verifies the ability to perform genetic association studies. The odds ratios between 1.  [20] and a recent meta-analysis (OR 1.21 and 1.35, depending on age) [21], confirming the high discriminatory power of our approach. Assuming that the finally available number of subjects will be .6,000, that number would be adequate for genetic analysis to detect low to intermediate effects (OR 1.2) for common SNPs (MAF.20%). For analysis of very low effects (OR,1.2) or less frequent variants (MAF,10%) pooling with other studies providing similar phenotypes will be necessary (Table S2). The Leipzig (LIFE) Heart Study has already been used for such a strategy [22].
Another advantage is the availability of cohorts with different entities of CAD. This also allowed analyses within CAD populations. Interestingly, we found that patients with LMCAD had a stronger association with chromosome 9p21 than patients with obstructive CAD without LMCAD. This finding is supported by studies displaying a higher heritability of left main and proximal CAD [23,24]. In contrast, comparing subjects with stable obstructive CAD to subjects with myocardial infarction, no association with 9p21 variants was seen.  The latter finding is supported by previous work, showing that 9p21 did not associate with myocardial infarction once stratified for disease severity [25]. Even though these results show the general suitability of the study for analyses of CAD subtypes, larger sample sizes will be needed to obtain more robust results.
In conclusion, the design and recruitment mode of the Leipzig (LIFE) Heart Study provides a promising basis for innovative molecular and diagnostic studies in patients with coronary artery disease. The identification of new potential targets for lifestyle and pharmacological interventions may lead to an improvement of prevention, diagnosis and treatment of atherosclerotic diseases.  Data S1 Supporting information with a detailed description of methods. (DOCX)